The effect of psychosocial and healthcare interventions on hospitalisation in people with dementia: an umbrella review

ObjectiveTo evaluate the efficacy of psychosocial and healthcare interventions on reducing hospitalisation in people with dementia (PwD). DesignUmbrella review of existing systematic reviews and meta-analyses. Search strategyMEDLINE, Embase, Cochrane, and CINAHL were searched until September 2025 for relevant keywords and medical subject headings. Inclusion criteriaPeer-reviewed systematic reviews with or without meta-analyses examining the effect of any psychosocial or healthcare intervention compared to a suitable control group on any measure of hospitalisation in people with dementia. MethodsTwo reviewers assessed eligibility, and eligible papers were grouped into categories and appraised by two reviewers using AMSTAR (A MeaSurement Tool to Assess systematic Reviews) version 2. We assessed certainty of evidence for each intervention type using GRADE (Grading of Recommendations Assessment, Development, and Evaluation). ResultsThe search identified 25 systematic reviews, comprising 77 unique primary studies (47 randomised control trials, 30 non-randomised studies of interventions), totalling 1,483,077 participants. There was high-certainty evidence that case management and exercise programmes had no effect on hospitalisation of PwD. There was low certainty evidence that advance care planning (ACP) reduced hospital admissions. Moderate certainty evidence from one study suggested that including clinical pharmacists in multidisciplinary teams (MDTs) reduced medication-related hospital admissions. ConclusionsThe current evidence for psychosocial and healthcare interventions in reducing hospitalisation in dementia is insufficient to make strong recommendations. ACPs show promise in reducing hospitalisations, and clinical pharmacists in MDTs may reduce medication-related readmissions. As evidence-based recommendations are needed to reduce the burden of dementia-related hospitalisations, more robust research is needed, especially high-quality randomised control trials with greater detail and standardised checklists for interventions and outcomes to reduce heterogeneity and establish more confident clinical recommendations. RegistrationPROSPERO 2024 CRD42024604296